Role permissions
Roles follow your org chart — owners, GCs, subs, suppliers, finance, safety, field, admin.
Skip to main content
Security & trust
Governed construction intelligence starts with trust
Ezelogs is built on role permissions, tenant boundaries, audit trails, and governed AI — recommendations are explainable and human-approved.
No third-party attestation, federal authorization, or health-data compliance claim is made on this page unless independently verified.
Trust posture
How Ezelogs treats trust
Trust is a function of permissions, boundaries, evidence, and human control — not a marketing badge.
Tenant boundaries
Company, project, and partner data live behind tenant boundaries with no cross-tenant blending.
Auditability
Signal → recommendation → human decision → action is captured and reviewable.
Governed AI
AI proposes; humans approve. High-risk workflows require explicit approval.
Human approval
No autonomous-action workflows. Approvals always live with the responsible human role.
Evidence-backed recommendations
Every recommendation surfaces the underlying records it was derived from.
Governance boundaries
What is separated, and how
Ezelogs treats company, project, role, AI, and admin boundaries as distinct surfaces — not overlapping pools.
CompanyProjectRoleAIAdmin
Company data
Tenant-bounded by default. Cross-tenant sharing only via explicit, audited agreements.
Project data
Project access mirrors the project participants — owner, GC, subs, suppliers, AEC, consultants.
Role access
Roles control what users can read, edit, approve, and export.
AI actions
AI proposes recommendations; it does not execute high-risk actions without human approval.
Admin controls
Admin actions are scoped, logged, and reviewable — including configuration changes.
AI governance
How AI behaves inside Ezelogs
AI is a participant inside a governed workflow — not an autonomous operator.
AI proposes
Recommendations are surfaced to the responsible role, not executed silently.
Humans approve
High-impact actions require explicit human approval before they take effect.
Explainable
Recommendations show the signals and records they were derived from.
Auditable
Every recommendation, approval, and action is captured for review.
High-risk gating
Cost, schedule, contractual, and safety-impact actions are gated behind review.
Tenant controls
Conservative multi-tenant posture
Tenants are bounded. No cross-tenant blending in recommendations or analytics surfaces.
Logical tenant isolation
Tenant identifiers scope every read, write, and recommendation.
Tenant-scoped credentials
API credentials and integrations are scoped to a tenant, not shared globally.
Explicit cross-tenant sharing
Cross-tenant collaboration (e.g., owner ↔ GC ↔ sub) is opt-in, scoped, and audited.
No cross-tenant model training
Tenant data is not used to train cross-tenant models without explicit, written consent.
Auditability evidence trail
Signal → recommendation → human decision → audit trail
Every governed recommendation can be traced from its source signal to the human who approved it.
1
Signal
Operational signal observed (RFI, cost, schedule, observation, submittal).
2
Recommendation
AI proposes an explainable recommendation tied back to the underlying records.
3
Human decision
The responsible role reviews, approves, modifies, or declines the recommendation.
4
Audit trail
The signal, recommendation, decision, and resulting action are recorded for review.
Security proof rows
Claim → evidence → status
Security statements should map to operating practice or independent verification — not marketing language.
Security posture statements describe current operating practice. Formal third-party attestations or certifications are listed only when independently verified.
Security FAQ
Common questions from procurement, IT, and security reviewers
Conservative answers. Anything that requires independent verification is flagged for Cursor review.
No. AI proposes recommendations; the responsible human role approves before action.
Role-aware access mirrors construction-industry org structure — owners, GCs, subs, suppliers, AEC, finance, safety, field, admin.
Each recommendation is explainable, traces back to its source records, and is captured in an audit trail.
Tenants are bounded by tenant identifiers across reads, writes, and recommendations. No cross-tenant blending.
High-risk workflows — cost, schedule, contractual, and safety-impact actions — require explicit human approval.
Any reference to third-party attestations, federal authorizations, or health-data compliance frameworks must be independently verified before public use.
Security review
Bring your security review questions. We will answer the way auditors expect.
Sandbox does not list certifications. Cursor wires verified attestations on merge.